What kind of news gatekeepers do we want machines to be? Filter bubbles, fragmentation, and the normative dimensions of algorithmic recommendations
作者：
Highlights：
• In real-world testing on a diverse sample, news recommendations were homogeneous.
• The top recommendations were consistently identical for conservatives and liberals.
• 41 publishers (99.9% of recommendations) reached both conservatives and liberals.
• On average, 69% of all recommendations were to five news organizations.
• The most-recommended five publishers accounted for 49% of the links collected.
摘要
•In real-world testing on a diverse sample, news recommendations were homogeneous.•The top recommendations were consistently identical for conservatives and liberals.•41 publishers (99.9% of recommendations) reached both conservatives and liberals.•On average, 69% of all recommendations were to five news organizations.•The most-recommended five publishers accounted for 49% of the links collected.
论文关键词：Algorithm,Echo chamber,Filter bubble,Fragmentation,Gatekeeping,Google News,Journalism,News diversity,Personalization
论文评审过程：Received 2 January 2018, Revised 5 June 2018, Accepted 29 July 2018, Available online 9 August 2018, Version of Record 18 November 2018.
论文官网地址：https://doi.org/10.1016/j.chb.2018.07.043
